Hey there! Talking about me , I like working with data and i think thats the most accurate and shortest summary of all my works . Talking about my works, most of my work lives somewhere between data science, machine learning, analytics, and data engineering. I enjoy starting from raw, unstructured data and slowly shaping it into something useful, whether that is a model, a dashboard, or a system someone can actually rely on.

I care a lot about things making sense.
Clear assumptions. Clean pipelines. Honest evaluation.

I am still learning, still experimenting, and still figuring out what “good” looks like in real-world systems, but that curiosity is what keeps me here.

Tech: PostgreSQL, Python, Power BI

• Built PostgreSQL analytics database processing 1,900+ streaming titles, implementing advanced SQL to normalize multi-value fields and compute 10+ KPIs across country, genre, runtime, and IMDb metrics.
• Analyzed cross-regional content trends revealing Korea’s 53.57% market dominance and 10× growth in yearly production since 2010, identifying high-rating niche genres to guide data-driven licensing strategy.
• Designed a Power BI dashboard with 10+ interactive, growth, ranking, and genre-rating views to support content acquisition and catalog optimization decisions.

🔗 Source Code

Tech: PostgreSQL, Python, Power BI

• Engineered PostgreSQL database processing 645M+ votes across 543 constituencies, designing 50+ advanced SQL queries using window functions and CTEs to analyze nationwide party performance patterns.
• Developed four analytical indices (Competitiveness, Efficiency, Fragmentation, Popularity), revealing BJP’s ˜62% vote-to-96% seat conversion in Gujarat and 20+ swing constituencies with less than 1% margins.
• Built interactive Power BI dashboards with 15+ DAX measures tracking postal ballot impact and margin analysis, delivering actionable insights for strategic resource allocation and electoral trend forecasting.

🔗 Source Code

Tech: AWS S3, AWS Glue, Athena, Power BI

• Built an end-to-end AWS-based data engineering pipeline using S3, Glue, Athena, and Power BI to process and analyze Spotify datasets.
• Transformed raw CSV data into analytics-ready Parquet tables using AWS Glue ETL jobs, applying data cleaning, normalization, and joins.
• Enabled efficient querying and reporting through Athena SQL and Power BI dashboards by optimizing storage formats and data structure.

🔗 Source Code

Tech: Python, ASR, Rasa NLU, FastAPI

• Built end-to-end voice banking pipeline supporting balance checks, transfers, and queries across 7 Indian languages via ASR, text normalization, NLU, and API routing .
• Engineered FastAPI backend with OTP verification, liveness detection, audit logging, and DPDP-compliant storage for secure financial transaction execution .
• Designed a scalable architecture with routing layers, mock DB, HSM-like signing & automated testing utilities, reducing pipeline failure rates by 40% .

🔗 Source Code

Tech: Python, LangChain, FAISS, Generative AI

• Built RAG-based document intelligence system using FAISS vector search and Gemini embeddings for semantic retrieval and natural-language querying across multi-PDF datasets
• Engineered end-to-end pipeline with PDF extraction, recursive chunking, and vector indexing, reducing irrelevant retrieval matches by 40%
• Developed interactive Streamlit interface with custom LangChain QA chain and prompt-engineered templates, improving response consistency by 30% over baseline LLM

🔗 Source Code

Tech: Python, NLP, Recommender Systems

• Developed a hybrid recommendation engine using Latent Dirichlet Allocation (LDA) & Cosine Similarity, achieving 90% accuracy & improving user engagement by 25% over baseline.
• Analyzed 100+ reviews and book descriptions to extract key themes and match user preferences effectively to develop a responsive UI that enables real-time recommendations.
• Integrated scikit-learn’s cosine similarity, resulting 35% increase in recommendation relevance over traditional collaborative filtering models.

🔗 Source Code

Tech: Flask, MySQL

• Developed a student-teacher portal & session management supporting 100+ logins & 2 access roles.
• Enabled grade upload/viewing workflows; teachers can assign grades to 50+ students across 10+ subjects, while students accessed only their own records.
• Secured user data with regex-based validations, session tokens & access checks preventing unauthorized access

🔗 Source Code

• Amazon Machine Learning Summer School 2025
• Cisco CCNA: Enterprise Networking Security and Automation
• Cisco CCNA: Introduction to Networks
• Cisco CCNA: Switching Routing and Wireless Essentials
• Coursera Advanced Learning Algorithms
• Coursera Generative AI: Introduction and Applications

• 📗 I read a LOT , like genuinely a LOTT
• 🎧 I love me some chatpata desi playlist for late night debugging sessions , controversial take but i dont like coffee so these high energy beats keep me awake.
• 🧶 I also crochet , and have a small crochet business